Re: [PATCH net 3/4] bonding: fix xfrm real_dev null pointer dereference

[Hangbin Liu <liuhangbin@gmail.com>]

On Mon, Aug 19, 2024 at 10:34:16AM +0300, Nikolay Aleksandrov wrote:
> On 19/08/2024 05:54, Hangbin Liu wrote:
> > On Fri, Aug 16, 2024 at 02:48:12PM +0300, Nikolay Aleksandrov wrote:
> >> We shouldn't set real_dev to NULL because packets can be in transit and
> >> xfrm might call xdo_dev_offload_ok() in parallel. All callbacks assume
> >> real_dev is set.
> >>
> >>  Example trace:
> >>  kernel: BUG: unable to handle page fault for address: 0000000000001030
> >>  kernel: bond0: (slave eni0np1): making interface the new active one
> >>  kernel: #PF: supervisor write access in kernel mode
> >>  kernel: #PF: error_code(0x0002) - not-present page
> >>  kernel: PGD 0 P4D 0
> >>  kernel: Oops: 0002 [#1] PREEMPT SMP
> >>  kernel: CPU: 4 PID: 2237 Comm: ping Not tainted 6.7.7+ #12
> >>  kernel: Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-2.fc40 04/01/2014
> >>  kernel: RIP: 0010:nsim_ipsec_offload_ok+0xc/0x20 [netdevsim]
> >>  kernel: bond0: (slave eni0np1): bond_ipsec_add_sa_all: failed to add SA
> > 
> > I saw the errors are during bond_ipsec_add_sa_all, which also
> > set ipsec->xs->xso.real_dev = NULL. Should we fix it there?
> > 
> > Thanks
> > Hangbin
> 
> Correct, I saw it too but I didn't remove it on purpose. I know it can lead to a
> similar error, but the fix is more complicated. I don't believe it's correct to
> set real_dev if the SA add failed, so we need to think about a different way
> to sync it. To be fair in real life it would be more difficult to hit it because
> the device must be in a state where the SA add fails, although it supports
> xfrm offload. The problem is that real_dev must be set before attempting the SA
> add in the first place.

Got it, so this time we only fix the delete path.

Reviewed-by: Hangbin Liu <liuhangbin@gmail.com>